Poker AI Again Trounces Human Challengers – This Time From China

Alan Du’s never met a poker adversary with such a stone-cold demeanor. It’s the fifth day that the venture capitalist and World Series of Poker veteran has gone up against this opponent – and the losses are stacking up.

His rival is literally inhuman, Du conceded after repeatedly keying bets into a computer and eliciting nary a hint of emotion. That’s because he went up against “Lengpudashi:” an updated version of the Libratus artificial intelligence program that achieved a major milestone by besting four of the world’s best poker pros in January.

Housed within a supercomputing center near Carnegie Mellon University in Pittsburgh, its name, intended to resemble its English moniker, fittingly translates into “cold poker master.”

Du and five team members played 36,000 hands against the machine over the course of five days. On Monday, at a resort conference center on China’s Hainan island, the final point-based score was announced: the AI won by a landslide.

Poker is a popular game among venture capitalists because “every hand you play is like a venture, trying to assess risk and ROI,” said Du, a seed investor who became the first mainland Chinese to win a WSOP gold bracelet in Las Vegas last year. “We held ourselves very well when playing against this world-class opponent.”

Poker’s complex betting strategies and the element of bluffing make it particularly intriguing to AI researchers. A player also decides to bet, bluff or fold without ever seeing the opponent’s full hand – a different kind of challenge than games like chess or Go, in which all the pieces are clearly visible on a playing board.

Du had tried to prevail where the pros had fallen short by employing an understanding of AI. Unlike the players in the January match-up who drew upon years of professional experience, Du’s Chinese team, which included a former Oracle engineer and startup entrepreneurs, attempted to apply their knowledge of machine intelligence and game theory to counter the machine’s moves. It wasn’t enough.

The latest AI exhibition, organised by Sinovation Ventures and Hainan’s government, didn’t generate quite the same buzz as last year’s match-up between Google DeepMind’s AlphaGo and Korean master Lee Sedol in Seoul. Perhaps that’s because even casual observers are becoming accustomed to seeing AI software upstage humans.

Tuomas Sandholm, a professor of computer science at Carnegie Mellon, has been honing the research underlying Libratus since 2004, honing its ability to make decisions in situations with imperfect information. The point of training AI to win at games like chess, Go, and poker isn’t for the sake of games themselves, but because controlled environments help computers hone strategic decision-making. Those reasoning skills can then be applied to real-world problems such as business, finance, and cybersecurity, he said.

“People have a misunderstanding of what computers and people are each good at. People think that bluffing is very human – it turns out that’s not true,” said Noam Brown, Sandholm’s PhD student and a co-developer of Libratus. “A computer can learn from experience that if it has a weak hand and it bluffs, it can make more money.”

The AI didn’t learn to bluff from mimicking successful human poker players, but from game theory. “Its strategies were computed from just the rules of the game,” not from analysing historical data, Sandholm said.

Venture capitalist Kai-Fu Lee, founder of Sinovation and an event organizer, said the rapid acceleration of AI technology over the past five years wasn’t possible before the advent of big data analysis. His fund has invested $120 million (roughly Rs. 775 crores) in AI-related companies in China – including facial recognition and loan-application startups – and he plans to devote a significant chunk of the money he’s currently raising to other AI ventures.

Also evident in the Hainan exhibition was the possibility of AI’s gradual democratisation. Brown said the computing power on display over the competition could be had for under $20,000 (roughly Rs. 12.9 lakhs).

“It’s surprisingly affordable,” he said. “Within 5 years, this could be running on smartphones.”

Capsule Full of Space Station Junk Makes Fiery Re-Entry

A capsule filled with space station trash bit the cosmic dust Wednesday with a keenly interested scientific audience.

The cargo carrier broke apart and burned up while re-entering Earth’s atmosphere high above the Pacific. Researchers gathered information on the breakup from sensors it carried, in hopes that the data will improve future spacecraft.

Capsule Full of Space Station Junk Makes Fiery Re-Entry

Built by Virginia-based Orbital ATK Inc., the capsule had been cut loose from the International Space Station last week. It followed its own orbit until it was steered into the atmosphere, where it was consumed by the heat of re-entry.

The vessel had been the scene of another fiery experiment shortly after it left the space station. A large blaze deliberately was ignited in it so Nasa could study the spread of flames in weightlessness. Like the re-entry test, this, too, was aimed at enhancing spacecraft safety.

The so-called “Cygnus” vessel was named the S.S. Rick Husband, after the commander of the doomed space shuttle Columbia. Husband and six other astronauts were killed as Columbia broke apart during re-entry in 2003, the result of launch damage.

Nasa is paying Orbital ATK, as well as SpaceX, to stockpile the space station. SpaceX has another station supply run coming up next month, while Orbital ATK plans to resume launches from Wallops Island, Virginia, in August. Orbital ATK temporarily moved its Cygnus flights to Cape Canaveral, Florida, after its Antares rocket exploded during liftoff from Wallops Island in 2014.

This latest Cygnus was launched to the space station from Florida back in March, using another company’s rocket. Once the supplies were unloaded, the capsule was filled with more than 4,000 pounds of garbage and discarded equipment for disposal. It had delivered twice that much.

Promising lab-grown skin sprouts hair and grows glands

Scientists in Japan have successfully transplanted mice with lab-grown skin that has more of the organ’s working parts in place than ever before.

Starting with stem cells made from a mouse’s gums, they managed to craft skin with multiple layers – as well as hair follicles and sweat glands.

When implanted into a “nude mouse” with a suppressed immune system, it integrated well and sprouted hairs.

Researchers say this success will take 5-10 years to translate into humans.

But eventually, the team hopes their system will lead to perfectly functioning skin that can be grown from the cells of burns victims and transplanted back on to them.

skin samples sprouting hair

Personalised organs

This would be vastly superior to the culturing and grafting techniques that are currently available, which produce skin without many of the the biological components and functionality that we are used to.

The technique could also be adapted to manufacture realistic skin samples that drug or cosmetics companies could use to test their products – instead of using animals.

The findings, reported in the journal Science Advances, have been greeted with enthusiasm by other scientists working in this field.

Takashi Tsuji is the paper’s senior author. He said the dream of re-growing personalised organs was beginning to materialise:

“Up until now, artificial skin development has been hampered by the fact that the skin lacked the important organs, such as hair follicles and exocrine glands, which allow the skin to play its important role in regulation.

“With this new technique, we have successfully grown skin that replicates the function of normal tissue.

“We are coming ever closer to the dream of being able to recreate actual organs in the lab for transplantation.”

Dr Tsuji, from the Riken Centre for Developmental Biology in Kobe, conducted the research with colleagues in Tokyo, Sagamihara and Sendai.

They began their experiments by taking cells from the gums of a mouse and converting them into “induced pluripotent stem cells” or iPSCs.

This is a popular and promising technique in stem cell research, discovered in 2006, which bathes the cells in chemicals to “wind back the clock”. The resulting cells, like those of an embryo, can divide again and again, and be guided down many developmental pathways to become nearly any type of cell in the body.

The team’s real achievement was in coaxing these cells to form the different layers and structures of deeply layered skin – the “integumentary organ” that protects our bodies, senses touch, regulates heat and does myriad other jobs as well.

‘Whole box of stuff’

John McGrath, a professor of molecular dermatology at King’s College London, said this study was one that researchers in his field had been looking out for – and it was a substantial step forward.

He told BBC News that the new system took us “over the halfway mark” towards growing functional skin for human patients – where previous efforts had stumbled at much earlier stages.

“It’s recapitulating normal skin architecture,” Prof McGrath said. “So rather than having isolated bits of skin… here we’ve actually got a whole box of stuff.

“To give you a football analogy: anybody can have Wayne Rooney, but now we’ve got Manchester United. There’s a whole team on the pitch, of interacting players.”

And that means there is hope, he added, for lifelike, lab-grown skin.

“[Today’s skin grafts] function, but they don’t really look like or behave like skin. If you don’t have the hair follicles and you don’t have the sweat glands and things, it’s not going to function as skin.”

Prof McGrath also said that many other laboratories would now be trying to reproduce these findings – and to adapt them for different purposes, such as recreating skin diseases in a dish and trying out treatments.

“There will be lots of benefits for immediate use, as well as for translational science,” he said.

Ocean heating doubles

The ocean is taking heat. That’s the conclusion of a new study that finds that Earth’s oceans now absorb heat at twice the rate they did 18 years ago. Around half of ocean heat uptake since 1865 has taken place since 1997, researchers also report online January 18 in Nature Climate Change.

coral reef

Warming waters are known to contribute to coral bleaching (SN Online: 10/8/15) and they take up more space than cooler waters, raising sea levels. While the top of the ocean is well studied, its depths are trickier to query. The researchers gathered 150 years of ocean temperature data in order to get a better picture of heat uptake from surface to seabed. They compiled temperature readings collected by everything from a 19th century sailing expedition of theHMS Challenger to modern automated ocean probes. The far-flung data sources, combined with computer simulations, created a timeline of ocean temperature changes, including cooling from volcanic eruptions and warming from fossil fuel emissions.

About 35 percent of the heat taken in by the oceans during the industrial era now resides at a depth of more than 700 meters, the researchers found. They say they’re unsure whether the deep-sea warming offset warming at the sea’s surface.

Time running out on comet lander

Time is running out for the Philae comet lander. The latest attempts to communicate with the probe have failed. And with temperatures plummeting as the comet races from the sun, it will soon be too cold for the robotic explorer to keep its computer running.

On January 10, engineers sent a command to spin Philae’s reaction wheel, which helped stabilize the lander during its descent. They hoped to nudge the lander into sunlight or at least shake several months of comet dust off its solar panels. Philae did not respond.

“It’s a bit sad,” says Stephan Ulamec, Philae’s manager at the German Aerospace Center in Cologne. “But it would be sad if we concentrated so much on what we couldn’t achieve and not on what we did.”

Philae’s mission got off to a rough start (SN: 12/13/14, p. 6). After separating from the Rosetta spacecrafton November 12, 2014, Philae bounced across comet 67P/Churyumov-Gerasimenko and settled against cliffs where there was not enough sunlight to keep the power on. Philae spent just 55 hours investigating its new home before shutting down. As the comet moved along its orbit, Philae’s solar panels spent more time in the sun. Seven months after going quiet, Philae phoned home after its batteries recharged (SN Online: 6/14/15). Contact with the lander has since been intermittent, with its last communication — a 12-minute burst of data from its radar instrument.

Philae

“The situation gets worse every day,” says Ulamec. “There’s little reason to believe that if the lander doesn’t wake now, it will wake in worse conditions in a few weeks.” As the comet travels farther from the sun, less solar power is available to the lander. By the end of January, temperatures will have dropped below −51° Celsius, Ulamec says, at which point the computers will no longer boot up.

“The fact that it worked at all is miraculous,” says Jessica Sunshine, a planetary scientist at University of Maryland in College Park. Before Rosetta launched in 2004, researchers knew very little about what comets are like up close. Landing on a comet “was a gutsy thing,” she says.

During its brief active tenure on the comet, Philae got the first intimate pictures of any comet and detected a fog of organic compounds. Its radar found that 67P is porous and uniform throughout. “That was a fantastic measurement,” says Sunshine. The lack of layers in the comet’s interior suggests that 67P was put together gently, which means the nucleus is a time capsule carting around a preserved sample of detritus from which the solar system formed 4.6 billion years ago (SN: 8/22/15, p. 13).

Philae and Rosetta showed that features both tiny and enormous look similar on the comet, says Sunshine. Without context, it’s hard to distinguish between the rugged terrain around the lander and the cliffs towering about 900 meters over the comet’s midsection. “That’s telling us something about how this comet was put together and evolved,” she says. “These datasets are going to stand the test of time.”

Philae might be done exploring, but it won’t be forgotten. Ulamec hopes to get some images of Philae this summer as Rosetta cozies up to get a closer look at how 67P changed during its closest approach to the sun. At the end of the mission in September (SN Online: 6/23/15), engineers will direct Rosetta to crash on the comet, snapping pictures all the way down. “It won’t be a proper landing,” says Ulamec. But at least Philae will finally have some company.

Self-Filling Water Bottle Converts Humid Air into Drinkable H2O

When water is scarce, why not pull it out of thin air? An industrial designer in Austria is hoping to do just that (well, sort of).

Kristof Retezár, a designer based in Vienna, invented a device that can extract humidity from the air and condense it into drinkable water. The handy gadget, dubbed Fontus, can be attached to a bike so that cyclists can generate water during long-distance rides through the countryside, where pit stops may be few and far between.

Fontus works using the basic principle of condensation, which can be easily demonstrated by taking something out of a refrigerator (for instance, a can of soda) and leaving it on the kitchen counter for a bit. Eventually, you’ll notice moisture collecting on the sides of the object.

“This is simply condensation of the humidity that is contained in the air,” Retezár told Live Science. “You always have a certain percentage of humidity in the air, it doesn’t matter where you are — even in the desert. That means you would always potentially be able to extract that humidity from the air.”

Fontus Self-Filling Water Bottle

The solar-powered device consists of a condensator (which functions like a cooler) that is connected to a series of hydrophobic surfaces that repel water. As the bike-mounted gadget takes in air, and these surfaces get cold, you’re left with condensation, Retezár said.

“Because they’re hydrophobic, they immediately repel the condensed water that they created, so you get a drop flow [into the bottle],” he explained. “Basically, you’re taking air in a vapor state and converting it into a liquid state.”

Fontus can produce 0.5 quarts (0.5 liters) of water in 1 hour in what is considered “really good” conditions, with temperatures between 86 degrees and 104 degrees Fahrenheit (30 to 40 degrees Celsius) and between 80 percent and 90 percent humidity, Retezár said.

The prototype includes a filter at the top to keep dust and bugs out of the water, but currently it does not include a way to filter out potentially harmful contaminants.

“The water you get is clean, unless the air is really contaminated,” Retezár said. “We’re thinking about making a bottle that also has a carbon filter, and this one would be for cities or areas where you might think the air is contaminated. But originally, this water bottle was thought to be used in nature, and places where you wouldn’t have contaminated air.”

Retezár is also working on a stand-alone version that uses an inverted ventilator to suck air into the system (rather than relying on the airstream created from a moving bike). This next-generation version could be used in regions of the world where humidity is high, but water is scarce.

“The idea was to solve a global problem: water issues in areas of the world where there is very little groundwater but very high humidity,” Retezár said. “My intent was to invent a machine or device that would be able to filter the humidity in the air and turn it into drinkable water.”

The initial Fontus design was shortlisted for the 2014 James Dyson Award, which helped Retezár gain exposure for the project, he said. Since then, he has received funding from the Austrian government that will help cover the technical development phase. The designer is also aiming to launch a crowdfunding campaign in March to cover the cost of mass-producing the devices. Retezár said he is aiming to keep the retail price for the Fontus under $100, and if all goes according to plan, the self-filling bottles could be commercially available in about nine or 10 months.